I used to work part-time in 1st level IT support in a local hospital when I was younger.
The main "theme" of my superior's work subjects there (2009-2016) was the migration from XP to 7. You heard that right.
And apart from the usual Office- and AD-Lock-In, the most problematic workstations of course were always ones with very specialized software. Virtualization and terminal services were in use, but the whole selling point of Windows was mostly put ad absurdum already, because they needed Windows licenses for dedicated machines running e.g. specialized MRT software, but those weren't even part of the main network anyway. They needed arcane syncing procedures anyway and Windows provided no value whatsoever on these devices. Same for the patient monitoring systems on ICU beds. These were using some "embedded" Windows and were rarely working in a stable way at all, nor way they connected to the networks running AD or the CIS (edit: seems it's called HIS in English)
CAD and stuff in the office divisions was similar, but with less integration needs (apart from network printing)
What I'm trying to say is: like in many offices, any slight change made users hostile, updates cost obscene amounts of work and money, and Windows didn't provide much more value compared to SAMBA. That is dated experience, I know.
But MS has not shown to be a trustworthy company in any of my work experience so far.
It was impossible to create working solutions without MS, yes, but the reasons for that never seemed to be grounded in actual value provided by an MS-centric software and networking structure.
It was just the one available commercial solution with enough adoption, and MS has been milking their target markets with these strategies for a very long time.
Making themselves "indispensable", even in machines where their software was used to run a terminal server, basically.
Hell, in my town, 3 years ago, they started to replace subway train LED signals with crappy Windows-CE-based software.
The effects are still noticeable... the whole infrastructure is still 80% worse compared to 10 years ago.
You recognize the useless Windows licenses by the occasional Desktop (seriously, google "cologne KVB windows trashcan"....), 90deg-tilted display, and of course 20% of the signage is out of operation on average now.
I think the long-lasting solution will be to move to a web-based application system, instead of depending on Desktop applications made for Windows or Linux. Using a web app system, the government only has to concern itself with proper development and maintenance of servers and web apps, and the public workers can use any operating system with a web browser.
Zig is trying to fix C, by delivering stuff already present in Modula-2, minus comptime, in a C like clothing.
The judge is still out there if it will become something unavoidable in some industry circles, or just yet another programming language to talk about.
From my point of view, being stuck in the past of systems languages design, during the AI age where the actual language becomes slowly irrelevant, it will mean the latter.